The complete set of these minutes and attachments is available in Portable Document Format (PDF) on the MODIS Home Page at http://modarch.gsfc.nasa.gov/MODIS/MODIS.html.
Introduction
The MODIS Science Team met on October 22-24, 1997, at the Holiday Inn in College Park, MD. Vince Salomonson, MODIS Team Leader, convened the meeting and welcomed participants. He expressed concern that MODIS is a bit behind the other teams whose instruments will fly on the AM-1 platform, with regard to algorithm preparation, etc., and added that at this meeting he hopes that the Science Data Support Team (SDST) will clarify exactly what deliveries it needs, and when. Turning to the Protoflight Model (PFM) instrument, Salomonson reported that it is on the spacecraft and is scheduled to enter thermal vacuum testing in December. Finally, he announced that some of the newly-selected validation scientists are in attendance today, and welcomed them to the meeting.
MODIS Characterization Support Team (MCST) Presentations
MCST made two presentations to the MST: "Characterization of MODIS in the VIS/NIR and SWIR" (on October 22) and "Characterization of MODIS in the MWIR and LWIR" (on October 24). Bruce Guenther outlined MCST's two primary objectives: first, to provide MST members all they need to know to understand the instrument and accomplish their goals for Level 2 products and above; and second, to acquire feedback from MST members on issues such as the handling of SWIR features in algorithms, as well as what approach to take with the linear or nonlinear algorithm decision.
MCST Summary
Before summarizing the MCST presentations Guenther announced that, during the first instrument comprehensive performance test, telemetry flowed successfully from Valley Forge to the MCST computer facility at GSFC. He reminded MST members that the MCST look-up tables (LUT) use the Santa Barbara Remote Sensing (SBRS) detector-numbering convention, which is inverted from the pixel convention used in the MODIS Level 1 products. Guenther outlined Level 1B file format changes, noting that at this meeting he had hoped to get an MST recommendation as to how to handle the SWIR 500-meter bands "second-sample" problem.
Guenther reviewed the major PFM issues still to be resolved. The stray light issue (On-board Calibrator-Blackbody [OBC-B], for high scan angles, +50 degrees and higher) will be investigated further pending results from an improved FM-1 test. With regard to SWIR radiometric behavior, Guenther cited two issues: the spectral leak, which is now being studied, and the second-sample problem, which might possibly be solved with an algorithm fix on-orbit. MCST is still waiting for access to the Spherical Integrator Source (SIS), so as to make round-robin measurements; this must be done before March 1998. Salomonson concluded that overall we have a robust instrument; we know what it can and cannot do, and we should expect good results.
Data Production and Readiness Plans
Masuoka reviewed the Version 1 Product Generation Executives (PGE) that have been integrated at the GSFC DAAC (GDAAC), EROS Data Center (EDC) and National Snow and Ice Data Center (NSIDC). He discussed the lessons learned during Version 1, including resolution of environmental issues (i.e., compatibility of things like compilers, toolkits, and standards checkers between the Team Leader Computer Facility [TLCF] and the DAAC).
Steve Wharton presented an overview of the GDAAC Version 2 Science Software Integration & Test (SSI&T) schedule. He reported that Version 1 SSI&T will run through mid-November, after which Version 2 SSI&T will formally begin. He indicated that workarounds are being explored so that more PGEs can be ready in time for launch. He added that the Version 2 SSI&T schedule is much more aggressive than the Version 1 schedule; lessons learned in Version 1, improved coordination between SDST and GDAAC, and the availability of necessary hardware should expedite the Version 2 SSI&T process. Wharton stated that the GDAAC hopes to have 16 PGEs ready for certification testing by April 15, and another 11 PGEs are expected to be ready at launch. Murphy noted that timely code delivery is essential to the GDAAC's maintaining its schedule.
Jeff Eidenshink discussed the EDC Version 1 and 2 SSI&T schedule. He reported that a total of 6 Version 1 PGEs have been delivered so far, and the average infusion time per PGE is 3 weeks. He anticipates that 9 of 31 Version 2 PGEs (all MODIS) should be delivered pre-launch.
Spencer Shiotani reported that NSIDC has completed SSI&T for PGEs 43 (Daily Snow Cover) and 45 (8-day Gridded Snow Cover). Both PGEs have metadata problems that will be deferred to Version 2. Infusion testing is underway for PGE 44 (Daily Sea Ice). PGE 47 (8-day Gridded Sea Ice) has not yet been received.
Masuoka reviewed the Version 2 SSI&T timeline and pointed out that the launch-critical release must be at the DAAC by January 15. The at-launch system must be certified by May 30. Certification testing will encompass 3 days of continuous processing and, among other things, focus on time transitions, terminator crossing, crossing the poles, and handling errors. Turning to the Version 2 schedule, Masuoka reported that many developers have not delivered their code when promised. Late code deliveries ripple through the whole system and can jeopardize other products in a chain.
Network Status
Masuoka summarized network status for Sol Broder. Masuoka announced that the ESDIS network budget will be scrubbed to save money. Broder asks that MST members consider how the downsizing of product volumes will impact the size of Quality Assurance (Q/A) volumes requested.
MODIS Emergency Backup System (MEBS)
Bill Engelmeyer reported that the MEBS has either completed, or is on target to complete, all of its prototyping milestones. MEBS is intended to be ready at launch for some limited processing. MEBS science products can be ordered via the Web at http://ltpwww.gsfc.nasa.gov/MODIS/SDST/mebs.
At-launch Volumes and Loads
Masuoka reviewed the 25, 50, 75, 100% processing ramp-up recommended by the EOS Review Group (ERG) to reduce costs without serious impact to EOS science. He noted that there are more resources at the DAACs than are necessary to satisfy the phased processing requirement at launch. Masuoka called attention to the proposed allocation of resources at the GDAAC. Of the 3,120 total MFLOPS available, 1,507 are available for higher-level products (i.e., higher than Level 1). Based on the June 1997 baseline, higher-level processing at the GDAAC could possibly be divided as follows: MODLAND, 609 MFLOPS; MOCEAN, 701 MFLOPS; and Atmosphere (without Cloud Mask), 197 MFLOPS. Masuoka asked that each of the discipline groups consider whether this strawman allocation of resources at the GDAAC will work and, if not, how should resources be allocated? Al Fleig asked that the discipline groups also consider how they will assist SDST in debugging immediately after launch, as well as how best to make and distribute early images (i.e., "first light products"). Once all of the groups have answered these questions, it will be necessary to ensure that their respective plans mesh so that MODIS presents a unified position to EOS.
Discipline Group Responses to SDST Questionnaire
MODLAND
At-launch Resource Allocation: Chris Justice indicated that MODLAND would like to test algorithms with a global month of Level 1B data; once the algorithms are tested and verified, then MODLAND would like to make products on a regional basis in order to satisfy the 25% constraint. MODLAND would like to run all of its products operationally at full resolution at a Year+1 and have sufficient resources to support reprocessing of these data.
At-launch Code Problem Resolution Approach: Justice suggested that MODLAND needs to work more as a team with SDST and the GDAAC. Overall, there needs to be a more hands-on approach where the discipline group members can assist in implementing adjustments to code.
Early Products: MODLAND proposes to use the MEBS strings being developed and, through the Land Data Operational Product Evaluation (LDOPE) Facility, run the early products there. Justice proposed that MODIS package up data sets for the community at large, and EDC could distribute these packages via CD or WWW.
MOCEAN
At-launch Resource Allocation: Wayne Esaias reported that MOCEAN plans to make all of its Level 2 products at 1-km resolution, based on hardware available at the GDAAC and MODIS processor allocations; if necessary to meet a 25% constraint, only every other pixel will be used. As more processing comes online, MOCEAN would like to go back and start reprocessing 1 year after launch. Esaias added that a T-1 link exists between Miami and GSFC, although it will have to be utilized to full capacity. The T-3 link planned in the ESDIS baseline must be installed if the Miami Ocean Science Computing Facility (SCF) is to play an active role in the Q/A of ocean data and refining MOCEAN algorithms post-launch.
At-launch Code Problem Resolution Approach: MOCEAN will perform debugging remotely from Miami.
Early Products: MOCEAN believes that it can make products in a matter of days, provided that good Level 1B data are available. Esaias indicated that having Cloud Mask would be helpful, but it is not absolutely necessary. Fluorescence, chlorophyll, and productivity for the visible bands are all important products to disseminate quickly.
Atmosphere
At-launch Resource Allocation: Gumley reported that Atmosphere would meet the 25% constraint by processing data from roughly 8 days a month. UW is concerned that the T-1 line currently in place may not be adequate to get all the necessary Level 1B data, but if the University of Wisconsin (UW) becomes part of the Very-High-Performance Backbone Network Service (VBNS), then the requisite bandwidth will be available. Fleig asked if MOCEAN had any problems with the 8-day Atmosphere plan, and Esaias said that it did not. (Cloud Mask will be produced everywhere all the time anyway.)
At-launch Code Problem Resolution Approach: Atmosphere already has a good working relationship with Rich Hucek and SDST, and there is no need for major changes between now and launch. Much of the Atmosphere group is already on-site at GSFC, and debugging can be done remotely from UW as well.
Early Products: Atmosphere would prefer to produce "first light" images independently.
SDST Summary
Masuoka outlined MODIS PGE delivery priorities, explaining that Priority 1 PGEs include Level 1 products (3), PGEs for EOS Ground System (EGS) certification (15), and at-launch PGEs (19). Priority 2 PGEs are post-launch PGEs, of which there are 27. More PGEs may be added to the post-launch list, depending on deliveries from the MST, the speed with which the SDST and DAAC personnel can get PGEs integrated and tested at the DAACs, and the impact of external requirements changes on our software delivery. Masuoka stressed that, for at-launch PGEs, the software must meet standards and be robust. The schedule for integration is tight, and few Version 2.1 PGEs can be accepted without dropping at-launch PGEs. PGEs that are delivered late may end up post-launch, or cause other PGEs to be moved to post-launch. Masuoka requested that the discipline groups assist SDST as much as possible with the SSI&T process after delivery, whether this means having somebody on-site or on-call to help fix identified problems. In order to better monitor PGE status, Michael King requested that Masuoka make his PGE tracking spreadsheets available via the Web, and Masuoka agreed to do this.
MODIS Visualization Tool
Dave Santek demonstrated the MODIS Visualization Internet Environment - Wisconsin (MODVIEW) tool. MODVIEW will be Web-based, and enable the user to browse large databases, display MODIS data, and overlay maps, latitude/longitude lines, and ancillary data. More information is available via the Web at http://www.ssec.wisc.edu.
MODIS Cloud Mask Status
Steve Ackerman stated that he intends to address Cloud Mask validation, and focus on visualization of results, "ground truth" (Lidar [Cloud Lidar System (CLS)] and weather observations), users, and intercomparisons with other instruments. Gumley demonstrated the prototype visualization tool, called Sharp, developed at UW using MODIS Airborne Simulator (MAS) data. The software is very user-friendly, and runs on any platform that runs IDL. Among other things, the tool allows the user to add the cloud mask overlay, do some simple band math, and create Graphic Image Format (GIF) files on screen. The software is available on the Web at http://cimss.ssec.wisc.edu/~gumley/sharp/sharp.html. Ackerman presented a series of graphs showing that MAS Cloud Mask results agree with Lidar data. MAS Cloud Mask also compares favorably with AVHRR data. Ackerman reported that MODIS and MISR personnel are now investigating how best to compare data at launch.
ESDIS Status
John Dalton summarized ESDIS Status and discussed the August ECS demo as it pertains to MODIS. He reported that ESDIS is planning for an incremental approach to ECS releases, in order to provide earlier access to high-priority functions and better serve evolving community needs. Dalton noted that the Release B.0', B.0, B.1 series of releases has been replaced by Version 2.0, 2.1, etc.; he reviewed the timeline showing when ESDIS expects to have each version ready. (Version 2.0 is due in June 1998, and Version 2.1 is due in November 1998). Turning to the August demo, Dalton indicated that, overall, it was a success. The original criteria consisted of 46 functions; of these, 3 involved production rules that were not exercised by the available PGEs. MODIS PGEs 1, 2, and 8 were used to demonstrate chaining. Dalton acknowledged that there is much work to be done to tune the ECS system to meet at-launch performance requirements. Performance bottlenecks and tuning steps have been identified, and a plan is in place to support the AM-1 and Landsat-7 data flows. In reviewing the phased development of ECS capabilities to meet at-launch needs, Dalton reported that some automated operations such as system fail-over may be introduced after launch and, therefore, increase the need for manual operations procedures early in the mission. Asked to define certification, Dalton responded that certification means that the DAACs have the operating elements of the system, as well as some PGEs they can use to test and verify that the system does what it is supposed to do. Certification does not mean that all of the at-launch science software is in the system and is fully tested; rather, it is a pre-launch demonstration that the DAAC operators and science software developers can use the system to perform essential functions. Dalton addressed the land tiling issue, stating that the basic capability to gather granules within a tile will be in pre-launch, operationally-tested Drop 3. However, the capability to cluster tiles to make best use of resources will not be integrated until near launch, in Drop 4.
Global Imager (GLI) Status
Hiroshi Murakami presented a status report on NASDA's GLI Science Mission. He explained that GLI is a general purpose, medium-spatial-resolution visible/infrared imager that will make atmosphere, land, and ocean-color observations. GLI is scheduled to be launched in 1999.
AM-1 Status
Ken Anderson reported that the PFM instrument will go into thermal vacuum testing in early- to-mid-December, and it will remain there approximately five weeks. Following thermal vacuum testing, the instrument will be shipped to the West Coast in plenty of time for launch. Anderson announced that there are presently no major concerns that would prevent launch as scheduled in June 1998.
PM-1 Status
Marty Donohoe presented an overview of PM-1 status. Overall the project is in good shape as regards technical and schedule issues, although cost reserves have been reduced. The PM-1 Project will be discussing Integration and Test (I&T) and flight operations with TRW in November. Donohoe indicated that MODIS FM-1 instrument issues now being addressed include the SWIR light leak, test flow modifications, and scan mirror replacement. He thanked Anderson for ensuring that MODIS instrument issues are addressed quickly by SBRS. Donohoe noted that MODLAND had been concerned about FM-1 stability and pointing; he believes that this issue has been resolved and summarized in a memo, but he will double-check this. Claire Parkinson presented an update on the algorithms for the two instruments (AMSR and AIRS) aboard PM-1, but not AM-1. MST members may want to think in terms of collaborative efforts with these two instrument teams.
MOCEAN Splinter Session Summary
Esaias reported that the new validation affiliates are meshing nicely, and he anticipates that they will make major contributions. The Marine Optical Buoy (MOBY) array is deployed and operational, and being used to good effect for SeaWiFS. A SeaWiFS initialization cruise is scheduled for January 10 through February 8. Funds for the MODIS initialization cruise must be obligated within months in order to assure ship availability. Information on MOBY/Marine Optical Characterization Experiment (MOCE) activities and data is available online at: http://moby.mlml.calstate. edu. With regard to the instrument, Esaias indicated that he is investigating the potential benefits of replacing the scan mirror; so far it appears that replacement will provide substantially more useful data near clouds. MOCEAN was pleased to learn that the deep space maneuver has been baselined. With regard to products, Esaias reported that Version 2.0 code will be delivered on schedule; he anticipates that delivery of Version 2.1 code will be on schedule as well, although this schedule remains TBD. Turning to resource allocation, he reiterated that MOCEAN plans to make all of its Level 2 products at 1-km resolution, based on hardware available at the GDAAC and MODIS processor allocations; if necessary to meet a 25% constraint, only every other pixel will be used. As more processing comes online, MOCEAN would like to go back and start reprocessing 1 year after launch.
Atmosphere Splinter Session Summary
King voiced Atmosphere's concern that crosstalk between SWIR bands needs to be better characterized. Guenther has agreed to provide additional analysis, and Atmosphere algorithm developers will assess impacts. Also, the crosstalk in Bands 33 - 36, discussed earlier by Moeller, will have a significant impact on Atmosphere algorithms. King indicated that Atmosphere is very concerned about the subsampling problem, whereby Bands 5 and 7 are well characterized for different pixels than Band 6. He reviewed the dates when Atmosphere code is expected to be delivered to SDST. He noted that execution of Level 3 code requires Version 7.2 of the Fortran compiler, which the GDAAC does not now support. Masuoka has been apprised of this issue, and it is being worked. Finally King reported that a Kalahari desert validation campaign is planned for August/September 1999. This will be a collaborative effort with MODLAND and others.
MODLAND Splinter Session Summary
Alan Strahler reported that MODLAND is concerned about the light leaks affecting Bands 6 and 7, but there are no show-stoppers as regards the instrument. MODLAND would like to see faster SDST turnaround on code adjustments; after delivery to SDST, MODLAND developers will chaperone their products through the system. With regard to production reduction, Strahler expressed concern that EOS science is in jeopardy if peer-reviewed products are cut. MODLAND feels that the proposed ramp-up of EOSDIS capacity is too slow. There is the need for full operational capacity to generate the suite of MODLAND global products in near real-time at Launch+1 year, and to start reprocessing those products which have undergone significant refinement, completing the reprocessing by Launch+2 years. Strahler reviewed MODLAND's strategy to meet the proposed ramp-up. As regards early MODIS products, MODLAND recommends land data packages, which would be multi-instrument packages designed to meet general user needs in the first 9 months. (These packages would be coordinated through the Science Working Group for the AM Platform [SWAMP].) Strahler indicated that MODLAND is concerned about network capacity, and would like reassurance that plans will be implemented and DAAC-SCF links will be tested prior to launch (6 months lead time). Finally, Strahler reported that MODLAND feels more confident as regards geolocation. However, MODLAND would like to see a plan for post-launch implementation and a schedule for prototyping/testing the land control-point algorithm, which has slipped from earlier expectations.
Closing Remarks
Salomonson thanked the newly-designated validation scientists for attending the meeting. He also thanked Barbara Conboy and the MODIS Administrative Support Team (MAST) for coordinating the logistics. Salomonson stated that the instrument appears to be in good shape; it is not perfect, but we are aware of the few flaws and are dealing with them accordingly. The next step for the MST is data products, and much remains to be done in this area. Once code is delivered and PGEs are integrated, the Team will turn its attention to validation and science. The next meeting is tentatively scheduled for May 1998 in the GSFC area.